Is being data driven a fad?

Tell me the odds now, but maybe not next year.

How much should Trump Media & Technology Group, the parent company of Truth Social, be worth?

One way to answer that question is to carefully study its quarterly filings, use key performance and operating metrics, like product signups and active users and advertising revenue, to create various mathematical models that forecast its future cash flows, adjust those estimates to account for various probabilistic risks, develop residual income models and dividend discount models, and smash it all together into a giant equation full of Greek letters and LaTeX formulas.¹ And then, just as an astrophysicist can say where Jupiter’s moons will be in April of 2030, you can say what Trump Media & Technology Group should be worth today. 

A business school professor would probably recommend an approach like this. But Trump Media & Technology Group would not:

Investors should be aware that since its inception, TMTG has not relied on any specific key performance metric to make business or operating decisions. Consequently, it has not been maintaining internal controls and procedures for periodically collecting such information, if any. While many mature industry peers may gather and analyze certain metrics, given the early development stage of the Truth Social platform, TMTG’s management team believes that such metrics are not critical in the near future for the business and operation of the platform. …

TMTG will consider the relevant key performance indicators for its then-current business operations and determine whether it has effective controls and procedures in place to process information related to the disclosure of key performance indicators and metrics. Should this be the case, TMTG may decide to collect and report such metrics if they are deemed to significantly enhance investors’ understanding of TMTG’s financial condition, cash flows, and other aspects of its financial performance. However, TMTG may find it challenging or cost-prohibitive to implement such effective controls and procedures and may never collect, monitor, or report any or certain key operating metrics. 

In other words: Do not value our business based on models and math and numbers, because we don’t use the numbers, we don’t know the numbers, we don’t track the numbers, and we definitely don’t audit the numbers. This whole company is a bit; you know it’s a bit; we know it’s a bit; and we are required to tell the SEC, in obtuse legalese about risk factors and the cost-prohibitive nature of internal controls and procedures, that it is a bit. It’s a meme that trades on the Nasdaq; it’s a presidential election prediction market; it’s a commemorative collectable coin, sold as a soft con. And you should value it the way you would value those things: By betting on what you think other people will think it’s worth. The answer to what TMTG should be worth isn’t found in a math problem, but in—and only in—our animal spirits. 

Clearly, the market is using this second method to value TMTG. In the first quarter of 2024, TMTG made $770,500 in revenue, which is, one, about as much as a single McDonald’s location makes in a quarter, and, two, down from $1.1 million in the first quarter of 2023. To bring in that revenue, TMTG spent $99 million on operating expenses, for a net profit of negative $98 million.² 

As of this morning, the market values TMTG—a magic box that can, every three months, turn $99 million into $770,500, and declining—at $5.93 billion. For comparison, the market values the actual Box.com, which brought in $270 million in revenues, spent $250 million on expenses, and made $20 million of profit, at $4.63 billion. 

It is possible that this phenomenon is unique to TMTG. Maybe TMTG is attracting lots of politically-motivated investors who want to win Donald Trump’s favor by propping up his shell company. Maybe people think that Donald Trump will soon be the president, and he will reward government contracts to Truth Social, or require all federal agencies to post exclusively on Truth Social. Maybe they think Elon Musk will soon be president, and he will buy Truth Social for Twitter. Maybe the market views TMTG as a crypto token of Donald Trump himself, and maybe that is fun to own.³

More generally, maybe TMTG is a fad, and it will eventually return to its textbook value.

But maybe not. Maybe the textbook is the fad, and the true baseline of what a company should be worth isn’t based on dividends or residual income, but on the collective id. From Matt Levine:

I have written and thought a lot about meme stocks over the last few years, and it has made me a bit nihilistic. Sometimes I suggest that perhaps fundamental analysis—the idea that the prices of financial assets reflect the present value of their future cash flows—is a temporary phenomenon; people were trading stocks based on vibes and gamesmanship long before anyone built discounted cash flow models, and perhaps they’ll keep doing so long after. I mostly think of that as a social phenomenon: The available information about companies keeps getting better, but there is no law of nature that forces people to pay attention to that information. You can just buy GameStop because your friends are buying GameStop.

This is both an obvious point—maybe stocks are priced by vibes, and hype, and the general interest that people seem to have in buying stocks—and a profoundly hard one to actually accept. All the money and math has to mean something. Sure, maybe there are bouts of irrational exuberance, but there has to be some sort of fundamental bedrock underneath it all, right? It can’t just be a perpetual motion machine of popularity, can it? 

It…could be? It…probably is? Different methods for how to value stocks come in and out of vogue, and what is widely accepted in one decade can become outdated in the next. Over a longer evolutionary cycle, is it not possible for equity models to move beyond using financial indicators? If the only true anchor for a stock’s price is what other people will pay for it, are vibes a worse measure of its “fundamental value” than recurring revenue?⁴ A company’s hype factor may not be a red flag; it might be its only ground truth. 

---

My brief history of the modern analytics movement goes something like this:

Thirty years ago, people used to make decisions with a little bit of data. They would develop business strategies by looking at monthly sales reports, or roster moves by looking at tables of home runs and RBIs, or trade stocks by playing connect the dots on a line chart. This worked ok.

Then, fifteen or so years ago, a handful of companies, sports teams, and now-celebrities like Nate Silver became very rich and successful by doing more careful analysis. This worked better, in part because they were clever, in part because they were applying these techniques to problems for which data was particularly useful,⁵ and partly because their competition—other companies, other teams, pundits hand-counting yard signs—was immature.

We all saw this, and got very excited. Data became important; “being data-driven” became urgent. We started trying to quantify and optimize: A/B test everything; analyze this; analyze that. A cottage industry of content and a booming industry of business applications got built on the idea that everything in the future will become more scientific and more automated. Data literacy will become as important as actual literacy; every company will become a data company. 

And it kinda didn’t work? Or, at best, the results have been mixed. There are success stories—Wall Street is dominated by quant funds, for example—but there have also been lots of busts. Increasingly, “we are data-driven” feels less like a competitive advantage, and more like an empty sales pitch.⁶

In Silicon Valley—where so much of this began, and maybe where it will end?—you can feel the mood shifting. Though the foil to Paul Graham’s founder mode is manager mode, it could’ve just as easily been data mode. Founder mode is making decisions, moving quickly, being bold, and taking action based on what you know is right; data mode is obsessive measurement, running the experiment longer, being tentative, and abdicating ownership to the numbers. Similarly, today’s folk heroes embody the opposite of data-driven ideals: Elon Musk didn’t optimize Twitter; he went scorched earth on it. Sam Altman didn’t raise $6.6 billion by promising to moneyball a market; he did it by promising a grand utopia. And data and technology is no longer eating the world; taste is:

Just as software ate the world and transformed industries in the last era, taste is now eating software—and with it, Silicon Valley.

In this new era of Silicon Valley, taste isn’t just an advantage—it’s the future. The most compelling startups will be those that marry great tech with great taste. Even the pursuit of unlocking technological breakthroughs must be done with taste and cultural resonance in mind, not just for the sake of the technology itself. 

None of these loose philosophies are in outright opposition to being data driven, and there aren’t many explicitly data skeptics like I thought there might be. But there are fashionable alternatives. And the question, I suppose, is if they are just trendy wiggles along our inevitable march towards an increasingly quantified world, or if the whole data-driven thing was the fad.

---

When we imagine futuristic societies, we tend to picture them wrapped in some space age aesthetic, where everything is automated by giant data-crunching computers. In Iron Man, Tony Stark’s Jarvis can solve international crimes in three minutes. In Star Wars, C-3PO can calculate the odds of successfully navigating an asteroid in real time. In Star Trek, the nearly omniscient robot is literally named Data. 

I suspect that these movies, and hundreds others like them, are part of why we’re subconsciously attached to the idea that a stock must have a fundamentally sound price, and that the long arc of the universe is bending toward a world run on numbers. If the distant future is Hollywood’s flavor of science fiction, the near future probably includes better financial models and more corporate analytics. 

But as Matt Levine suggests, just because the computers can tell us the odds doesn’t mean we have to pay attention to them. Other methods could become more popular. After all, the scene from Star Wars isn’t famous because C-3PO knew the odds; it’s famous because Han Solo ignored them. 

1. Who are we kidding, it’s in Excel and the entire formula is just NPV.

2. If a burn multiple under 1 is great, and under 2 is good, over 3 is bad, what’s a burn multiple of 127? (And even this is too low! Burn multiples aren’t revenue divided by burn, but added revenue divided burn. In Q1 of 2023, TMTG made $1.1 million in revenue; in Q1 of 2024, they made $770,000. They spent $99 million in early 2024 to lose $330,000 in revenue. So their burn multiple is...negative? Infinity? An asymptote, rocketing into the beyond?) 

3. A Horcrux, if you will.

4. Though these things aren’t entirely mutually exclusive. When everyone uses the same financial model to value a company, be it a model about cash flow, or dividends, or the blockchain, or what people on Reddit think is funny, that model will be correct—not because it reflects any sort of capital-T truth, but because the model will determine the vibes. If we all agree that the best way to value a company is using its discounted cash flow, does discounted cash flow define the “true” value of the company, or does the implicit social agreement about cash flow define it? 

5. For example, if you’re looking for undervalued baseball players, there’s a relatively direct way to find them in a database of 12 million Major League Baseball at-bats. But if you were a fashion company looking to decide what to include in your next season’s collection, data’s a lot less useful. 

6. I am sorry, but if you are telling people that you can rank the effectiveness of a political ad “down to a tenth of a percentage point of precision in multiple categories,” you aren’t learning something; you are selling something.